which ordained that none but the palaces of the gods should be built of marble or stone—though it is strange that the saintly Asoka's disregard of the rule was not apparently imitated by any of his successors. But it must be remembered that the temple itself was an important adjunct of the royal palace—probably in Vedic times it was the palace itself and the citadel which the Aryan warriors used as their inner line of defence. The cella was the sacrificial chamber of the king, and the mandapam his durbar-hall. Its walls, as we see in the Kailāsa at Ellora, were battlemented and surrounded by chambers which might have been used by the king and his retinue before priestly tradition converted them into monasteries. Its gateways were the guard-rooms for the royal sentinels. The sun-windows which pierced the sides of the temple tower suggest loopholes for the archers of the royal bodyguard. This use of the temples as royal fortresses may partly account for the ruthlessness with which they were destroyed by the Muhammadan invaders. It would also explain why, until gunpowder was generally used in warfare, Indian builders did not find it necessary to use stone for the innumerable accessory buildings of a royal palace—wood and brick sufficed for all practical purposes, and were much more convenient materials. When, however, about the fourteenth century, they began to follow the Muhammadan custom of building the royal residence and the royal chapel separately, the one was built as solidly as the other; and from that time the palaces of Hindu kings provide a distinct and brilliant chapter in Indian architectural history. They are of far greater artistic interest than those of the Great Moguls, which were planned on similar lines and built by the same class of builders, the royal craftsmen of India, though
